1w darmowym pakiecie R

Projekt „Nowa oferta edukacyjna Uniwersytetu Wrocławskiego odpowiedzią na współczesne potrzeby rynku pracy i gospodarki opartej na wiedzy”

> summary(model) Call:

lm(formula = RETCAP ~ . - OBS, data = fad) Residuals:

Min 1Q Median 3Q Max

-0.164562 -0.024625 0.003809 0.019430 0.127693 Coefficients:

Estimate Std. Error t value Pr(>|t|) (Intercept) 0.29215 0.16253 1.797 0.0835 . WCFTCL 0.13931 0.24065 0.579 0.5675 WCFTDT 0.40819 0.33844 1.206 0.2382 GEARRAT 0.03072 0.11938 0.257 0.7989 LOGSALE 0.18166 0.16791 1.082 0.2888 LOGASST -0.18951 0.16330 -1.160 0.2560 NFATAST -0.16514 0.16928 -0.976 0.3379 CAPINT -0.01610 0.03343 -0.482 0.6339 FATTOT -0.10211 0.10519 -0.971 0.3403 INVTAST -0.21645 0.22283 -0.971 0.3400 PAYOUT -0.02086 0.01755 -1.189 0.2450 QUIKRAT -0.08285 0.10904 -0.760 0.4539 CURRAT 0.02000 0.07691 0.260 0.7968 ---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 0.06377 on 27 degrees of freedom Multiple R-squared: 0.8529, Adjusted R-squared: 0.7875 F-statistic: 13.04 on 12 and 27 DF, p-value: 2.783e-08

> modelAIC <- stepAIC(model,trace = F)

> summary(modelAIC) Call:

lm(formula = RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + INVTAST + PAYOUT + QUIKRAT, data = fad)

Residuals:

Min 1Q Median 3Q Max

-0.169005 -0.027080 0.002032 0.023350 0.135828 Coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.25188 0.10419 2.418 0.021495 * WCFTDT 0.51823 0.04172 12.422 8.78e-14 ***

LOGSALE 0.10059 0.05944 1.692 0.100268 LOGASST -0.10382 0.05997 -1.731 0.093015 . NFATAST -0.28260 0.07597 -3.720 0.000764 ***

INVTAST -0.16284 0.10308 -1.580 0.123995 PAYOUT -0.02429 0.01487 -1.633 0.112271

QUIKRAT -0.04696 0.02803 -1.675 0.103586 ---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 0.06026 on 32 degrees of freedom Multiple R-squared: 0.8443, Adjusted R-squared: 0.8102 F-statistic: 24.79 on 7 and 32 DF, p-value: 3.201e-11

> anova(model,modelAIC) Analysis of Variance Table

Model 1: RETCAP ~ (OBS + WCFTCL + WCFTDT + GEARRAT + LOGSALE + LOGASST + NFATAST + CAPINT + FATTOT + INVTAST + PAYOUT + QUIKRAT +

CURRAT) - OBS

Model 2: RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + INVTAST + PAYOUT + QUIKRAT

Res.Df RSS Df Sum of Sq F Pr(>F) 1 27 0.10978

2 32 0.11618 -5 -0.0064013 0.3149 0.8997

> modelleaps<-regsubsets(RETCAP~.-OBS,data=fad,nbest=3)

> summary(modelleaps) Subset selection object

Call: regsubsets.formula(RETCAP ~ . - OBS, data = fad, nbest = 3) 12 Variables (and intercept)

Forced in Forced out WCFTCL FALSE FALSE WCFTDT FALSE FALSE GEARRAT FALSE FALSE LOGSALE FALSE FALSE LOGASST FALSE FALSE NFATAST FALSE FALSE CAPINT FALSE FALSE FATTOT FALSE FALSE INVTAST FALSE FALSE PAYOUT FALSE FALSE QUIKRAT FALSE FALSE CURRAT FALSE FALSE 3 subsets of each size up to 8 Selection Algorithm: exhaustive

WCFTCL WCFTDT GEARRAT LOGSALE LOGASST NFATAST CAPINT FATTOT INVTAST PAYOUT 1 ( 1 ) " " "*" " " " " " " " " " " " " " " " "

1 ( 2 ) "*" " " " " " " " " " " " " " " " " " "

1 ( 3 ) " " " " "*" " " " " " " " " " " " " " "

2 ( 1 ) " " "*" " " " " " " "*" " " " " " " " "

2 ( 2 ) " " "*" " " " " " " " " " " "*" " " " "

2 ( 3 ) " " "*" " " " " " " " " "*" " " " " " "

3 ( 1 ) " " "*" " " " " " " "*" " " " " " " " "

3 ( 2 ) " " "*" " " " " " " "*" " " " " " " " "

3 ( 3 ) " " "*" " " " " " " " " " " "*" " " " "

4 ( 1 ) " " "*" " " " " " " "*" " " " " " " "*"

4 ( 2 ) " " "*" " " " " " " "*" "*" " " " " " "

4 ( 3 ) " " "*" " " " " " " "*" " " " " " " "*"

5 ( 1 ) " " "*" " " " " " " "*" "*" " " " " "*"

5 ( 2 ) "*" "*" " " " " " " "*" " " " " " " "*"

5 ( 3 ) " " "*" " " " " " " "*" " " "*" " " "*"

6 ( 1 ) " " "*" " " " " " " "*" "*" " " "*" "*"

6 ( 2 ) " " "*" " " "*" "*" "*" " " " " " " "*"

6 ( 3 ) "*" "*" " " " " " " "*" "*" " " " " "*"

7 ( 1 ) " " "*" " " "*" "*" "*" " " " " "*" "*"

7 ( 2 ) "*" "*" " " "*" "*" " " " " "*" " " "*"

7 ( 3 ) "*" "*" " " " " " " "*" "*" "*" " " "*"

8 ( 1 ) " " "*" " " "*" "*" "*" " " "*" "*" "*"

8 ( 2 ) "*" "*" " " "*" "*" "*" " " " " "*" "*"

8 ( 3 ) "*" "*" " " "*" "*" "*" " " "*" " " "*"

QUIKRAT CURRAT 1 ( 1 ) " " " "

1 ( 2 ) " " " "

1 ( 3 ) " " " "

2 ( 1 ) " " " "

2 ( 2 ) " " " "

2 ( 3 ) " " " "

3 ( 1 ) " " "*"

3 ( 2 ) "*" " "

3 ( 3 ) " " "*"

4 ( 1 ) " " "*"

4 ( 2 ) " " "*"

4 ( 3 ) "*" " "

5 ( 1 ) " " "*"

5 ( 2 ) " " "*"

5 ( 3 ) " " "*"

6 ( 1 ) "*" " "

6 ( 2 ) " " "*"

6 ( 3 ) " " "*"

7 ( 1 ) "*" " "

7 ( 2 ) " " "*"

7 ( 3 ) " " "*"

8 ( 1 ) "*" " "

8 ( 2 ) "*" " "

8 ( 3 ) " " "*"

# plot a table of models showing variables in each model.

# models are ordered by the selection statistic.

> plot(modelleaps,scale="r2")

> summary(modelleaps1)

Subset selection object

Call: regsubsets.formula(RETCAP ~ . - OBS, data = fad) 12 Variables (and intercept)

Forced in Forced out WCFTCL FALSE FALSE WCFTDT FALSE FALSE GEARRAT FALSE FALSE LOGSALE FALSE FALSE LOGASST FALSE FALSE NFATAST FALSE FALSE CAPINT FALSE FALSE FATTOT FALSE FALSE INVTAST FALSE FALSE PAYOUT FALSE FALSE QUIKRAT FALSE FALSE CURRAT FALSE FALSE 1 subsets of each size up to 8 Selection Algorithm: exhaustive

WCFTCL WCFTDT GEARRAT LOGSALE LOGASST NFATAST CAPINT FATTOT INVTAST PAYOUT 1 ( 1 ) " " "*" " " " " " " " " " " " " " " " "

2 ( 1 ) " " "*" " " " " " " "*" " " " " " " " "

3 ( 1 ) " " "*" " " " " " " "*" " " " " " " " "

4 ( 1 ) " " "*" " " " " " " "*" " " " " " " "*"

5 ( 1 ) " " "*" " " " " " " "*" "*" " " " " "*"

6 ( 1 ) " " "*" " " " " " " "*" "*" " " "*" "*"

7 ( 1 ) " " "*" " " "*" "*" "*" " " " " "*" "*"

8 ( 1 ) " " "*" " " "*" "*" "*" " " "*" "*" "*"

QUIKRAT CURRAT 1 ( 1 ) " " " "

2 ( 1 ) " " " "

3 ( 1 ) " " "*"

4 ( 1 ) " " "*"

5 ( 1 ) " " "*"

6 ( 1 ) "*" " "

7 ( 1 ) "*" " "

8 ( 1 ) "*" " "

> plot(modelleaps1,scale="r2")

r2 (I n te rc e p t) W C F T C L W C F T D T G E A R R A T L O G S A L E L O G A S S T N F A T A S T C A P IN T F A T T O T IN V T A S T P A Y O U T Q U IK R A T C U R R A T

0.72 0.79 0.82 0.83 0.84 0.84 0.84 0.85

> coef(modelleaps1, 6:8) [[1]]

(Intercept) WCFTDT NFATAST CAPINT INVTAST PAYOUT QUIKRAT

0.22880281 0.51263547 -0.29308043 0.01807680 -0.15387173 -0.02180481 -0.05076139 [[2]]

(Intercept) WCFTDT LOGSALE LOGASST NFATAST INVTAST PAYOUT

0.25188458 0.51823378 0.10059360 -0.10382486 -0.28259719 -0.16283897 -0.02428761 QUIKRAT

-0.04696242 [[3]]

(Intercept) WCFTDT LOGSALE LOGASST NFATAST FATTOT INVTAST

0.25680768 0.51784783 0.10678311 -0.11086376 -0.19361242 -0.06699999 -0.16674265 PAYOUT QUIKRAT

-0.02300355 -0.04520680

> model8 <-

lm(RETCAP~WCFTDT+LOGSALE+LOGASST+NFATAST+FATTOT+INVTAST+PAYOUT+QUIKRAT)

> summary(model8) Call:

lm(formula = RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + FATTOT + INVTAST + PAYOUT + QUIKRAT)

Residuals:

Min 1Q Median 3Q Max

-0.168679 -0.028140 0.007812 0.022817 0.135031 Coefficients:

Estimate Std. Error t value Pr(>|t|) (Intercept) 0.25681 0.10516 2.442 0.0205 * WCFTDT 0.51785 0.04203 12.321 1.76e-13 ***

LOGSALE 0.10678 0.06046 1.766 0.0872 . LOGASST -0.11086 0.06116 -1.813 0.0796 . NFATAST -0.19361 0.14351 -1.349 0.1871 FATTOT -0.06700 0.09141 -0.733 0.4691 INVTAST -0.16674 0.10397 -1.604 0.1189 PAYOUT -0.02300 0.01508 -1.525 0.1374 QUIKRAT -0.04521 0.02834 -1.595 0.1208 ---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 0.0607 on 31 degrees of freedom Multiple R-squared: 0.847, Adjusted R-squared: 0.8075 F-statistic: 21.44 on 8 and 31 DF, p-value: 1.295e-10

> anova(modelAIC,model8) Analysis of Variance Table

Model 1: RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + INVTAST + PAYOUT + QUIKRAT

Model 2: RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + FATTOT + INVTAST + PAYOUT + QUIKRAT

Res.Df RSS Df Sum of Sq F Pr(>F) 1 32 0.11618

2 31 0.11420 1 0.0019793 0.5373 0.4691

> modelleapsb<-regsubsets(RETCAP~.-OBS,data=fad,method="backward")

> plot(modelleapsb,scale="r2")

r2 (I n te rc e p t) W C F T C L W C F T D T G E A R R A T L O G S A L E L O G A S S T N F A T A S T C A P IN T F A T T O T IN V T A S T P A Y O U T Q U IK R A T C U R R A T

0.72 0.79 0.81 0.82 0.82 0.83 0.84 0.85

> coef(modelleapsb,8)

(Intercept) WCFTDT LOGSALE LOGASST NFATAST FATTOT INVTAST

0.25680768 0.51784783 0.10678311 -0.11086376 -0.19361242 -0.06699999 -0.16674265 PAYOUT QUIKRAT

-0.02300355 -0.04520680

> modelleapsf<-regsubsets(RETCAP~.-OBS,data=fad,method="forward")

> plot(modelleapsf,scale="r2")

r2 (I n te rc e p t) W C F T C L W C F T D T G E A R R A T L O G S A L E L O G A S S T N F A T A S T C A P IN T F A T T O T IN V T A S T P A Y O U T Q U IK R A T C U R R A T

0.72 0.79 0.82 0.83 0.84 0.84 0.84 0.84

install.packages("relaimpo")

> library("relaimpo") lmg

is the R^2 contribution averaged over orderings among regressors last

is each variables contribution when included last, also sometimes called usefulness.

first

is each variables contribution when included first, which is just the squared covariance between y and the variable.

betasq

is the squared standardized coefficient.

relAIC<-calc.relimp(modelAIC,type=c("betasq","lmg","last","first"),rela=T)Response variable:

RETCAP

Total response variance: 0.01913327 Analysis based on 40 observations 7 Regressors:

WCFTDT LOGSALE LOGASST NFATAST INVTAST PAYOUT QUIKRAT Proportion of variance explained by model: 84.43%

Metrics are normalized to sum to 100% (rela=TRUE).

Relative importance metrics:

lmg last first betasq

WCFTDT 0.857597647 0.84794691 8.441544e-01 0.664451713 LOGSALE 0.025034641 0.01574205 4.056125e-02 0.109166563 LOGASST 0.028684624 0.01647378 5.917711e-02 0.111334872 NFATAST 0.041718855 0.07604036 6.176856e-06 0.077290238 INVTAST 0.020159534 0.01371490 2.920076e-02 0.014226852 PAYOUT 0.007503489 0.01465484 1.782710e-03 0.009327004 QUIKRAT 0.019301210 0.01542716 2.511761e-02 0.014202759

Average coefficients for different model sizes:

1X 2Xs 3Xs 4Xs 5Xs 6Xs

WCFTDT 0.436872590 0.447636717 0.46131502 0.476390495 0.49153843 0.50565906 LOGSALE 0.045859681 0.031338432 0.03243546 0.045314047 0.06414986 0.08364989 LOGASST 0.056612522 0.053767407 0.03251821 -0.001164189 -0.03932944 -0.07490490 NFATAST -0.001889466 -0.044925790 -0.09124368 -0.137498250 -0.18477081 -0.23405915 INVTAST -0.174482100 -0.140015024 -0.10793769 -0.089658920 -0.09418382 -0.12288292 PAYOUT -0.007941524 -0.008404619 -0.00961433 -0.012180865 -0.01603757 -0.02032435 QUIKRAT 0.046709279 0.033704143 0.01852123 0.002512363 -0.01390305 -0.03067533 7Xs

WCFTDT 0.51823378 LOGSALE 0.10059360 LOGASST -0.10382486 NFATAST -0.28259719 INVTAST -0.16283897 PAYOUT -0.02428761 QUIKRAT -0.04696242

> plot(relAIC)

WCFT LOGA INVT QUIK

Method LMG

% of R2 02060

WCFT LOGA INVT QUIK

Method Last

% of R2 02060

WCFT LOGA INVT QUIK

Method First

% of R2 02060

WCFT LOGA INVT QUIK

Method Betasq

% of R2 02060

Relative importances for RETCAP

R

 84.43%, metrics are normalized to sum 100%.

> lm.beta(modelAIC)

WCFTDT LOGSALE LOGASST NFATAST INVTAST PAYOUT QUIKRAT 1.0064042 0.4079299 -0.4119612 -0.3432441 -0.1472635 -0.1192372 -0.1471388

> summary(modelAIC) Call:

lm(formula = RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + INVTAST + PAYOUT + QUIKRAT, data = fad)

Residuals:

Min 1Q Median 3Q Max

-0.169005 -0.027080 0.002032 0.023350 0.135828 Coefficients:

Estimate Std. Error t value Pr(>|t|)

(Intercept) 0.25188 0.10419 2.418 0.021495 * WCFTDT 0.51823 0.04172 12.422 8.78e-14 ***

LOGSALE 0.10059 0.05944 1.692 0.100268 LOGASST -0.10382 0.05997 -1.731 0.093015 . NFATAST -0.28260 0.07597 -3.720 0.000764 ***

INVTAST -0.16284 0.10308 -1.580 0.123995 PAYOUT -0.02429 0.01487 -1.633 0.112271 QUIKRAT -0.04696 0.02803 -1.675 0.103586 ---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1 Residual standard error: 0.06026 on 32 degrees of freedom Multiple R-squared: 0.8443, Adjusted R-squared: 0.8102 F-statistic: 24.79 on 7 and 32 DF, p-value: 3.201e-11

>

> vif(modelAIC) # variance inflation factors

WCFTDT LOGSALE LOGASST NFATAST INVTAST PAYOUT QUIKRAT 1.349099 11.939261 11.635555 1.749966 1.785933 1.095746 1.585024

> sqrt(vif(modelAIC)) > 2 # problem?

WCFTDT LOGSALE LOGASST NFATAST INVTAST PAYOUT QUIKRAT FALSE TRUE TRUE FALSE FALSE FALSE FALSE

> (rdg<-select(lm.ridge(RETCAP ~ I(WCFTDT+0.25188)+ LOGSALE + LOGASST + NFATAST + + INVTAST + PAYOUT + QUIKRAT,data = fad,lambda = seq(0,0.1,0.0001))))

modified HKB estimator is 0.638 modified L-W estimator is 1.15 smallest value of GCV at 0.1

> lrdg<-lm.ridge(RETCAP ~ I(WCFTDT+0.25188)+ LOGSALE + LOGASST + NFATAST + + INVTAST + PAYOUT + QUIKRAT -1,data = fad,lambda = seq(0,0.1,0.01))

> lrdg$coef

0.00 0.01 0.02 0.03 0.04 0.05 0.06 0.07

I(WCFTDT + 0.25188) 0.2405 0.2388 0.2378 0.2370 0.2365 0.2359 0.2355 0.2351 LOGSALE 0.5647 0.4174 0.3353 0.2828 0.2464 0.2196 0.1990 0.1827 LOGASST -0.4719 -0.3246 -0.2425 -0.1903 -0.1541 -0.1276 -0.1074 -0.0915 NFATAST -0.0834 -0.0853 -0.0862 -0.0866 -0.0868 -0.0869 -0.0869 -0.0868 INVTAST -0.0362 -0.0301 -0.0266 -0.0243 -0.0227 -0.0214 -0.0204 -0.0196 PAYOUT -0.0186 -0.0172 -0.0164 -0.0159 -0.0156 -0.0153 -0.0151 -0.0150 QUIKRAT -0.0303 -0.0343 -0.0364 -0.0377 -0.0385 -0.0390 -0.0394 -0.0396 0.08 0.09 0.10

I(WCFTDT + 0.25188) 0.2347 0.2344 0.2341 LOGSALE 0.1695 0.1585 0.1493 LOGASST -0.0786 -0.0680 -0.0592 NFATAST -0.0867 -0.0866 -0.0864 INVTAST -0.0189 -0.0183 -0.0178 PAYOUT -0.0149 -0.0148 -0.0147 QUIKRAT -0.0398 -0.0399 -0.0399 ceresPlots(modelAIC)

-1.0 -0.5 0.0 0.5

-1.0-0.6-0.2

WCFTDT

CERES Residual(RETCAP)

4.0 4.5 5.0 5.5

-4.10-3.90

LOGSALE

CERES Residual(RETCAP)

3.5 4.0 4.5 5.0 5.5

-1.00-0.80

LOGASST

CERES Residual(RETCAP)

0.1 0.3 0.5 0.7

-0.40-0.25-0.10

NFATAST

CERES Residual(RETCAP)

0.0 0.1 0.2 0.3 0.4 0.5

-0.100.05

INVTAST

CERES Residual(RETCAP)

0 1 2 3 4

-0.60-0.45

PAYOUT

CERES Residual(RETCAP)

0.5 1.0 1.5 2.0 2.5

0.250.400.55

QUIKRAT

CERES Residual(RETCAP)

CERES Plots

> modelRobust <- rlm(formula = RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + + INVTAST + PAYOUT + QUIKRAT, data = fad)

> summary(modelRobust)

Call: rlm(formula = RETCAP ~ WCFTDT + LOGSALE + LOGASST + NFATAST + INVTAST + PAYOUT + QUIKRAT, data = fad)

Residuals:

Min 1Q Median 3Q Max

-0.163831 -0.020544 0.002268 0.020255 0.148438 Coefficients:

Value Std. Error t value (Intercept) 0.2310 0.0775 2.9811 WCFTDT 0.4984 0.0310 16.0662 LOGSALE 0.0892 0.0442 2.0182 LOGASST -0.0894 0.0446 -2.0058 NFATAST -0.2624 0.0565 -4.6440 INVTAST -0.1706 0.0766 -2.2254 PAYOUT -0.0239 0.0111 -2.1587 QUIKRAT -0.0394 0.0208 -1.8894

Residual standard error: 0.03047 on 32 degrees of freedom

> library(DAAG)

> CVlm(df=fad, modelAIC, m=3) # 3 krotna cross-validation Analysis of Variance Table

Response: RETCAP

Df Sum Sq Mean Sq F value Pr(>F) WCFTDT 1 0.537 0.537 147.93 1.5e-13 ***

LOGSALE 1 0.007 0.007 1.95 0.1719 LOGASST 1 0.026 0.026 7.02 0.0124 *

NFATAST 1 0.031 0.031 8.59 0.0062 **

INVTAST 1 0.005 0.005 1.45 0.2380 PAYOUT 1 0.014 0.014 3.77 0.0611 . QUIKRAT 1 0.010 0.010 2.81 0.1036 Residuals 32 0.116 0.004 ---

Signif. codes: 0 ‘***’ 0.001 ‘**’ 0.01 ‘*’ 0.05 ‘.’ 0.1 ‘ ’ 1

fold 1

Observations in test set: 11 12 13 14 16 22 23 25 26 34 35 37 39 X11 X12 X13 X14 X16 X22 X23 X25 X26

Predicted 0.21394 0.2521 2.00e-01 0.0952 0.3078 0.1728 0.169 0.1979 0.1210 RETCAP 0.21000 0.2200 2.00e-01 0.1100 0.2300 0.1700 -0.040 0.2100 0.1500 Residual -0.00394 -0.0321 -4.94e-05 0.0148 -0.0778 -0.0028 -0.209 0.0121 0.0290 X34 X35 X37 X39

Predicted -0.114 0.23 0.1881 0.1585 RETCAP 0.040 0.10 0.1000 0.1300 Residual 0.154 -0.13 -0.0881 -0.0285

Sum of squares = 0.1 Mean square = 0.0078 n = 13 fold 2

Observations in test set: 1 2 3 4 5 8 17 18 20 27 28 33 36 40 X1 X2 X3 X4 X5 X8 X17 X18 X20

Predicted 0.19237 0.1702 0.1577 0.1533 0.2437 0.1209 0.183 0.1449 0.1441 RETCAP 0.19000 0.2200 0.1700 0.1200 0.2100 0.1000 0.320 0.1300 0.0900 Residual -0.00237 0.0498 0.0123 -0.0333 -0.0337 -0.0209 0.137 -0.0149 -0.0541 X27 X28 X33 X36 X40

Predicted 0.1376 0.1736 0.0953 -0.1135 0.185 RETCAP 0.2300 0.2000 0.1400 -0.0900 0.080 Residual 0.0924 0.0264 0.0447 0.0235 -0.105

Sum of squares = 0.05 Mean square = 0.0036 n = 14 fold 3

Observations in test set: 6 7 9 10 15 19 21 24 29 30 31 32 38 X6 X7 X9 X10 X15 X19 X21 X24 X29 X30

Predicted 0.1696 0.1052 0.107 0.167 0.214 0.28313 -0.4750 0.2068 0.1322 0.0671 RETCAP 0.1200 0.1500 0.080 0.310 0.380 0.29000 -0.5000 0.2600 0.1900 0.0800 Residual -0.0496 0.0448 -0.027 0.143 0.166 0.00687 -0.0250 0.0532 0.0578 0.0129 X31 X32 X38

Predicted 0.1489 0.1626 0.19107 RETCAP 0.1900 0.2000 0.20000 Residual 0.0411 0.0374 0.00893

Sum of squares = 0.063 Mean square = 0.0049 n = 13 Overall ms

0.00536